yt_dlp/extractor/box.py

   1 import json
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     determine_ext,
   6     parse_iso8601,
   7     # try_get,
   8     update_url_query,
   9 )
  10
  11
  12 class BoxIE(InfoExtractor):
  13     _VALID_URL = r'https?://(?:[^.]+\.)?app\.box\.com/s/(?P<shared_name>[^/]+)/file/(?P<id>\d+)'
  14     _TEST = {
  15         'url': 'https://mlssoccer.app.box.com/s/0evd2o3e08l60lr4ygukepvnkord1o1x/file/510727257538',
  16         'md5': '1f81b2fd3960f38a40a3b8823e5fcd43',
  17         'info_dict': {
  18             'id': '510727257538',
  19             'ext': 'mp4',
  20             'title': 'Garber   St. Louis will be 28th MLS team  +scarving.mp4',
  21             'uploader': 'MLS Video',
  22             'timestamp': 1566320259,
  23             'upload_date': '20190820',
  24             'uploader_id': '235196876',
  25         }
  26     }
  27
  28     def _real_extract(self, url):
  29         shared_name, file_id = self._match_valid_url(url).groups()
  30         webpage = self._download_webpage(url, file_id)
  31         request_token = self._parse_json(self._search_regex(
  32             r'Box\.config\s*=\s*({.+?});', webpage,
  33             'Box config'), file_id)['requestToken']
  34         access_token = self._download_json(
  35             'https://app.box.com/app-api/enduserapp/elements/tokens', file_id,
  36             'Downloading token JSON metadata',
  37             data=json.dumps({'fileIDs': [file_id]}).encode(), headers={
  38                 'Content-Type': 'application/json',
  39                 'X-Request-Token': request_token,
  40                 'X-Box-EndUser-API': 'sharedName=' + shared_name,
  41             })[file_id]['read']
  42         shared_link = 'https://app.box.com/s/' + shared_name
  43         f = self._download_json(
  44             'https://api.box.com/2.0/files/' + file_id, file_id,
  45             'Downloading file JSON metadata', headers={
  46                 'Authorization': 'Bearer ' + access_token,
  47                 'BoxApi': 'shared_link=' + shared_link,
  48                 'X-Rep-Hints': '[dash]',  # TODO: extract `hls` formats
  49             }, query={
  50                 'fields': 'authenticated_download_url,created_at,created_by,description,extension,is_download_available,name,representations,size'
  51             })
  52         title = f['name']
  53
  54         query = {
  55             'access_token': access_token,
  56             'shared_link': shared_link
  57         }
  58
  59         formats = []
  60
  61         # for entry in (try_get(f, lambda x: x['representations']['entries'], list) or []):
  62         #     entry_url_template = try_get(
  63         #         entry, lambda x: x['content']['url_template'])
  64         #     if not entry_url_template:
  65         #         continue
  66         #     representation = entry.get('representation')
  67         #     if representation == 'dash':
  68         #         TODO: append query to every fragment URL
  69         #         formats.extend(self._extract_mpd_formats(
  70         #             entry_url_template.replace('{+asset_path}', 'manifest.mpd'),
  71         #             file_id, query=query))
  72
  73         authenticated_download_url = f.get('authenticated_download_url')
  74         if authenticated_download_url and f.get('is_download_available'):
  75             formats.append({
  76                 'ext': f.get('extension') or determine_ext(title),
  77                 'filesize': f.get('size'),
  78                 'format_id': 'download',
  79                 'url': update_url_query(authenticated_download_url, query),
  80             })
  81
  82         creator = f.get('created_by') or {}
  83
  84         return {
  85             'id': file_id,
  86             'title': title,
  87             'formats': formats,
  88             'description': f.get('description') or None,
  89             'uploader': creator.get('name'),
  90             'timestamp': parse_iso8601(f.get('created_at')),
  91             'uploader_id': creator.get('id'),
  92         }